Unintentional durotomy is a frequent complication of spinal surgical
procedures, with a rate as high as
17%1-5.
With appropriate management, the long-term outcomes for patients who had an
unintended durotomy were no different from those for patients who underwent a
similar surgery without
complications6,7.
A small percentage of patients with unintentional durotomy sustain
intracranial hemorrhage, likely due to an alteration in intradural hydrostatic
dynamics caused by cerebrospinal fluid
hypovolemia3,4,8,9.
Despite the gravity of this complication, to our knowledge, only six cases
have been previously
documented3,8-11.
All neurosurgeons and orthopaedic surgeons at a large tertiary care center
associated with a major university were solicited to identify cases of
intracranial hemorrhage following incidental durotomy that occurred during the
period of 1996 to 2006. The purpose of this report is to describe the four
identified patients who sustained intracranial hemorrhage following an
unintentional durotomy during posterior thora-columbar spinal surgery and to
discuss the importance of our findings. The patients were informed that data
concerning the case would be submitted for publication, and they
consented.
CASE 1. A seventy-seven-year-old man with a history of ankylosing
spondylitis and a daily intake of 325 mg of aspirin presented with a thoracic
kyphosis of 110° that impaired his ability to perform activities of daily
living. He was brought to the operating room for posterior arthrodesis and
instrumentation from T11 to S1 as well as L2 pedicle subtraction osteotomy.
During posterior dissection, a dural tear was made, resulting in a substantial
loss of cerebrospinal fluid. The dura was absent at the apex of the deformity
and was friable elsewhere. After an intraoperative neurosurgical consultation
was obtained, the dural rent was covered with DuraGen (Integra LifeSciences,
Plainsboro, New Jersey) and augmented with Tisseel (Baxter International,
Deerfield, Illinois) and an intradural subarachnoid drain was placed.
The patient was able to move the extremities postoperatively and was moved
to the neurosurgical intensive care unit. Over the next two hours, he became
increasingly somnolent and hemodynamically unstable even though the amount of
cerebrospinal fluid drainage was minimal. A computed tomographic scan of the
head three hours postoperatively indicated that the fourth ventricle was
filled with blood and that there was blood within the lateral ventricles and
among the cerebellar folia (Fig.
1). An intraventricular catheter was placed, and the condition of
the patient stabilized. Nine weeks after surgery, the patient was able to walk
with the aid of a walker but still had a major residual cognitive deficit. He
was discharged to a rehabilitation facility and died of aspiration pneumonia
nine months later.
Case 2. A fifty-five-year-old woman with a history of multiple
spinal operations presented with lumbar kyphoscoliosis and spinal canal
stenosis. She underwent a revision spinal arthrodesis from L2 to L4, anterior
corpectomy at L3, and an anterior-posterior reconstruction. Dissection at a
prior laminectomy site resulted in a 3-mm dural tear with an estimated
cerebrospinal fluid loss of <10 mL. The defect was primarily repaired and
was watertight to 40 cm of Valsalva pressure. The repair was not reinforced
with fibrin glue. A subfascial wound drain was placed and set to gravity
drainage without suction, and a daily average of approximately 235 mL of
serosanguineous fluid was collected before the drain was removed on the sixth
postoperative day. Motor function was intact postoperatively.
The patient was kept supine for the first postoperative day and then was
gradually mobilized. She had postural headaches on the third postoperative day
and expressive aphasia on the sixth postoperative day. A neurosurgical
consultation was obtained, and the patient was transferred to the
neurosurgical intensive care unit. A computed tomography scan of the head
revealed a subdural hematoma along the left temporal, parietal, and frontal
lobes. The hematoma, which had a maximal width of 1.2 cm and a 2-mm
left-to-right midline shift (Fig.
2), was evacuated via craniotomy, after which the cognition of the
patient normalized. On the tenth postoperative day, a serous leak was noted
from the distal aspect of the spinal surgical wound. The patient was once
again taken into surgery, and approximately 250 mL of serous fluid was noted
deep to the fascia at the site of the previous lumbar durotomy. The closure
was reinforced with Tisseel, and the patient was discharged without further
complications five days later.
Case 3. A sixty-three-year-old man with a history of type-II
diabetes mellitus and a daily intake of 325 mg of aspirin presented with
severe kyphoscoliosis and lumbar spinal stenosis. He underwent decompressive
laminectomies, posterior spinal osteotomies, and lumbosacral arthrodesis from
L2 to L5. A 6-mm durotomy occurred under the L2 lamina; it was repaired in a
watertight manner and reinforced with Tisseel. Two subfascial wound drains
were placed, and a daily average of 478 mL of serosanguineous fluid was
collected over a period of four days.
On the second postoperative day, the patient vacillated between
combativeness and somnolence and lost orientation to time and place and the
ability to open his eyes. A computed tomography scan of the head was acquired
forty-nine hours postoperatively and showed bilateral cerebellar subarachnoid
hematomas with mass effect and compression of the brainstem
(Fig. 3). He was transferred to
the neurosurgical critical care unit.
On the third postoperative day, the patient became fully alert and
oriented. The hematomas gradually resolved without any additional
interventions and the patient recovered completely and was discharged fifteen
days after surgery.
Case 4. A sixty-four-year-old woman presented with lumbar
scoliosis and spinal canal stenosis, for which she underwent decompressive
laminectomies and posterior spinal arthrodesis from L1 to the sacrum. During
laminectomy, a small dural tear occurred, which was primarily repaired and
tested with a Valsalva maneuver of 40 cm. Three subfascial wound drains were
placed, and a daily average of 660 mL of serosanguineous fluid was collected
over the next two days. She was able to move all of the extremities
postoperatively.
Warfarin therapy was initiated on the first postoperative day as
prophylaxis against venous thromboembolism. At 7 AM on the second
postoperative day, the patient complained of a severe headache. At 9 AM, she
was unresponsive and had a respiratory rate of six breaths per minute. She was
given 1 mg of naloxone intravenously and became verbally responsive, but was
unarousable ninety minutes later. Naloxone (1 mg) was administered
intravenously once again, and narcotics were discontinued. The patient became
more responsive. A neurosurgery consult took place at 3 PM when the patient
once again became unresponsive. Clinical examination revealed that the pupils
were 5 mm in diameter and were equal and nonreactive. The gag and corneal
reflexes were absent. A computed tomography scan of the head showed bilateral
cerebellar hemorrhage with obstructive hydrocephalus. An intraventricular
catheter was placed, but the patient did not improve and was declared brain
dead approximately forty-eight hours postoperatively.
Following this review, we were unable to identify consistent etiologic
factors for intracranial hemorrhage in these four patients. This complication
is rare but devastating and requires astute recognition of the early signs of
neurologic impairment. In addition to many nonspecific symptoms that can be
easily confused with postoperative delirium, which is especially common in
older patients12,
asymmetric or progressive neurologic deficits, cranial nerve involvement, and
aphasia suggest an intracranial abnormality.
The phenomenon of intracranial hemorrhage remote from a surgical site has
been described in the literature, and several authors have attempted to
explain the
pathophysiology8,13-15.
In 1981, Chadduck suggested that cerebrospinal fluid hypovolemia increases the
pressure gradient between intravascular and cerebrospinal fluids, thus causing
cerebellar veins to
rupture8.
Alternatively, the so-called sag model posits that hemorrhage results from
altered cerebrospinal fluid hydrodynamics, causing caudal sagging of the
cerebellum with stretching of cerebellar vermian
veins3,4,8,9,13,14.
In 1978, Moussa and Sharma postulated that rapid ventricular cerebrospinal
fluid loss causes the brain parenchyma to collapse, thereby causing stretching
and tearing of bridging
veins15. These
proposed mechanisms are not mutually exclusive. The model of rapid
cerebrospinal fluid loss that was described by Moussa and
Sharma15 occurred
in one of the patients in our study (Case 1). The sag model argues against the
early mobilization of patients who have undergone durotomy because standing
exacerbates the gravity-dependent cerebellar movement, which in turn causes
cerebellar
hemorrhage14. One
of the patients in our study (Case 2) was mobilized after the first
postoperative day, but symptoms of subdural hematoma did not begin to appear
until the third postoperative day. Furthermore, excessive drainage may cause
iatrogenic cerebrospinal fluid hypovolemia. Previous reports recommended that
cerebrospinal fluid drainage rates should be no more than 300 to 360 mL per
day, which is well below physiologic
production16-18.
The amount of subfascial drainage in two of the patients in our study (Case 3
and Case 4) exceeded these recommendations. We suggest that excessive drainage
should be prevented by occasionally clamping the drain(s).
We believe that intracranial hemorrhage following incidental durotomy is a
protean and insidious complication. Our series consists of two men and two
women with a mean age and standard deviation of 64.8 ± 9.0 years. While
revision surgeries have the greatest likelihood of being complicated by an
incidental
durotomy6, only one
of the four patients in our study was having a revision procedure. We are
aware of only one case of intracranial hemorrhage caused by incidental
durotomy following a revision procedure, and that case occurred in a patient
who had undergone prior
laminectomies11.
Three of the patients in our study used either antiplatelet or anticoagulation
medications preoperatively or postoperatively. Anticoagulation may be
contraindicated in the early postoperative period because the decreased
cerebrospinal fluid pressure cannot tamponade the epidural vessels. The dural
tears in our series were typically small; only the patient with ankylosing
spondylitis (Case 1) had a major defect due to friable dura, which is a
potential complication of that disease. All repairs, aside from the one in the
patient in Case 1, appeared to be watertight in the operating room, and all
patients awoke from surgery neurologically intact. An apparent watertight
intraoperative dural repair, however, does not preclude subsequent
cerebrospinal fluid leakage. It is unclear to us whether the use of fibrin
glue reinforcement decreases the risk of further leakage.
The symptoms experienced by the patient in Case 4 illustrate a crucial
point: nonspecific symptoms of postoperative neurologic deficit may mimic a
reaction to medication. Intracranial causes of altered mental status should be
considered in patients who had an incidental durotomy during surgery. Although
rare, intracranial hemorrhage following an incidental durotomy is a
devastating complication. Because intracranial hemorrhage can present with
variable symptoms, it is important that surgeons recognize and aggressively
evaluate early signs of neurologic deterioration to prevent a potentially
life-threatening complication. ?